1046 lines
28 KiB
JSON
1046 lines
28 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.4142502071251036,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.7426025390625,
|
|
"epoch": 0.004142502071251036,
|
|
"grad_norm": 4.96875,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7521,
|
|
"mean_token_accuracy": 9.009009227156639e-05,
|
|
"num_tokens": 9149.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742637062072754,
|
|
"epoch": 0.008285004142502071,
|
|
"grad_norm": 5.3125,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7074,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 16800.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742610836029053,
|
|
"epoch": 0.012427506213753107,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.743,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 26617.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742677021026612,
|
|
"epoch": 0.016570008285004142,
|
|
"grad_norm": 5.15625,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.6772,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 35855.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742655372619629,
|
|
"epoch": 0.020712510356255178,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.5975,
|
|
"mean_token_accuracy": 0.002011376223526895,
|
|
"num_tokens": 45644.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742458724975586,
|
|
"epoch": 0.024855012427506214,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4485,
|
|
"mean_token_accuracy": 0.019727057497948407,
|
|
"num_tokens": 54641.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741547870635987,
|
|
"epoch": 0.02899751449875725,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3138,
|
|
"mean_token_accuracy": 0.04090950228273869,
|
|
"num_tokens": 63924.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.739116668701172,
|
|
"epoch": 0.033140016570008285,
|
|
"grad_norm": 2.609375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.2108,
|
|
"mean_token_accuracy": 0.041662289202213286,
|
|
"num_tokens": 73452.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.735450172424317,
|
|
"epoch": 0.037282518641259324,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.0929,
|
|
"mean_token_accuracy": 0.039270054548978806,
|
|
"num_tokens": 82731.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.733051109313966,
|
|
"epoch": 0.041425020712510356,
|
|
"grad_norm": 2.171875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.0119,
|
|
"mean_token_accuracy": 0.04254167750477791,
|
|
"num_tokens": 92626.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.732081508636474,
|
|
"epoch": 0.045567522783761395,
|
|
"grad_norm": 1.9921875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.9713,
|
|
"mean_token_accuracy": 0.04115805197507143,
|
|
"num_tokens": 101380.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.730613136291504,
|
|
"epoch": 0.04971002485501243,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.8815,
|
|
"mean_token_accuracy": 0.041873040795326236,
|
|
"num_tokens": 109895.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.728298664093018,
|
|
"epoch": 0.053852526926263466,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.8626,
|
|
"mean_token_accuracy": 0.04000609517097473,
|
|
"num_tokens": 119203.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.7264967918396,
|
|
"epoch": 0.0579950289975145,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.7547,
|
|
"mean_token_accuracy": 0.03999764695763588,
|
|
"num_tokens": 127572.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.723791694641113,
|
|
"epoch": 0.06213753106876554,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.7222,
|
|
"mean_token_accuracy": 0.042604190111160276,
|
|
"num_tokens": 137161.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.719281482696534,
|
|
"epoch": 0.06628003314001657,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6617,
|
|
"mean_token_accuracy": 0.04084589965641498,
|
|
"num_tokens": 145991.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.712574768066407,
|
|
"epoch": 0.07042253521126761,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.598,
|
|
"mean_token_accuracy": 0.041171186976134776,
|
|
"num_tokens": 154927.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.706455898284911,
|
|
"epoch": 0.07456503728251865,
|
|
"grad_norm": 1.953125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.4593,
|
|
"mean_token_accuracy": 0.042302171140909194,
|
|
"num_tokens": 163234.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.699676418304444,
|
|
"epoch": 0.07870753935376967,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4018,
|
|
"mean_token_accuracy": 0.04330670088529587,
|
|
"num_tokens": 171554.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.688521003723144,
|
|
"epoch": 0.08285004142502071,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.3311,
|
|
"mean_token_accuracy": 0.047984727472066876,
|
|
"num_tokens": 180171.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.67230396270752,
|
|
"epoch": 0.08699254349627175,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.2155,
|
|
"mean_token_accuracy": 0.057312173396348955,
|
|
"num_tokens": 189216.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.645287895202637,
|
|
"epoch": 0.09113504556752279,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.2219,
|
|
"mean_token_accuracy": 0.052681009843945505,
|
|
"num_tokens": 200022.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.617320728302001,
|
|
"epoch": 0.09527754763877382,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.0422,
|
|
"mean_token_accuracy": 0.05372999124228954,
|
|
"num_tokens": 208686.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.564020252227783,
|
|
"epoch": 0.09942004971002485,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.9322,
|
|
"mean_token_accuracy": 0.054267140850424764,
|
|
"num_tokens": 218266.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.5111478805542,
|
|
"epoch": 0.1035625517812759,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.8288,
|
|
"mean_token_accuracy": 0.04845710247755051,
|
|
"num_tokens": 227658.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.454801654815673,
|
|
"epoch": 0.10770505385252693,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.6344,
|
|
"mean_token_accuracy": 0.06026250720024109,
|
|
"num_tokens": 236301.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.368635272979736,
|
|
"epoch": 0.11184755592377796,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.644,
|
|
"mean_token_accuracy": 0.048985954746603964,
|
|
"num_tokens": 246355.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.31434211730957,
|
|
"epoch": 0.115990057995029,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.4805,
|
|
"mean_token_accuracy": 0.05331834591925144,
|
|
"num_tokens": 256023.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.216940021514892,
|
|
"epoch": 0.12013256006628004,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.4309,
|
|
"mean_token_accuracy": 0.04547450765967369,
|
|
"num_tokens": 266893.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.09643030166626,
|
|
"epoch": 0.12427506213753108,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.1871,
|
|
"mean_token_accuracy": 0.0535514272749424,
|
|
"num_tokens": 276075.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.940307807922363,
|
|
"epoch": 0.12841756420878211,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.1018,
|
|
"mean_token_accuracy": 0.055915596336126326,
|
|
"num_tokens": 285529.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.774145030975342,
|
|
"epoch": 0.13256006628003314,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.0263,
|
|
"mean_token_accuracy": 0.05546179860830307,
|
|
"num_tokens": 294403.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.571882343292236,
|
|
"epoch": 0.13670256835128416,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.9661,
|
|
"mean_token_accuracy": 0.05883818194270134,
|
|
"num_tokens": 303507.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.380561351776123,
|
|
"epoch": 0.14084507042253522,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.7977,
|
|
"mean_token_accuracy": 0.06465772800147533,
|
|
"num_tokens": 312393.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.121123886108398,
|
|
"epoch": 0.14498757249378624,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.7071,
|
|
"mean_token_accuracy": 0.06362416185438632,
|
|
"num_tokens": 321309.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.919694709777833,
|
|
"epoch": 0.1491300745650373,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.7209,
|
|
"mean_token_accuracy": 0.06159159280359745,
|
|
"num_tokens": 330217.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.755216789245605,
|
|
"epoch": 0.15327257663628832,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.6795,
|
|
"mean_token_accuracy": 0.061438082903623584,
|
|
"num_tokens": 339587.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.5862530708313,
|
|
"epoch": 0.15741507870753935,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.5864,
|
|
"mean_token_accuracy": 0.06275138966739177,
|
|
"num_tokens": 349367.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.414569091796874,
|
|
"epoch": 0.1615575807787904,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.4852,
|
|
"mean_token_accuracy": 0.06507482491433621,
|
|
"num_tokens": 358272.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.34211015701294,
|
|
"epoch": 0.16570008285004142,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.6662,
|
|
"mean_token_accuracy": 0.06489905305206775,
|
|
"num_tokens": 367709.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.344364166259766,
|
|
"epoch": 0.16984258492129245,
|
|
"grad_norm": 0.87109375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.5038,
|
|
"mean_token_accuracy": 0.06977917924523354,
|
|
"num_tokens": 376180.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.172031211853028,
|
|
"epoch": 0.1739850869925435,
|
|
"grad_norm": 0.84765625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.6071,
|
|
"mean_token_accuracy": 0.06480413824319839,
|
|
"num_tokens": 385636.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.20558671951294,
|
|
"epoch": 0.17812758906379453,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.4616,
|
|
"mean_token_accuracy": 0.07232532948255539,
|
|
"num_tokens": 394486.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.116187858581544,
|
|
"epoch": 0.18227009113504558,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.4215,
|
|
"mean_token_accuracy": 0.06673903428018094,
|
|
"num_tokens": 403765.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.066840410232544,
|
|
"epoch": 0.1864125932062966,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.4037,
|
|
"mean_token_accuracy": 0.07279244847595692,
|
|
"num_tokens": 412988.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.047943782806396,
|
|
"epoch": 0.19055509527754763,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.3598,
|
|
"mean_token_accuracy": 0.07073828913271427,
|
|
"num_tokens": 422096.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.999078845977783,
|
|
"epoch": 0.19469759734879868,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.441,
|
|
"mean_token_accuracy": 0.06653320901095867,
|
|
"num_tokens": 431333.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.124928903579711,
|
|
"epoch": 0.1988400994200497,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.4798,
|
|
"mean_token_accuracy": 0.07552412785589695,
|
|
"num_tokens": 440382.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.967213916778564,
|
|
"epoch": 0.20298260149130073,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.3739,
|
|
"mean_token_accuracy": 0.07711177170276642,
|
|
"num_tokens": 448469.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.066874313354493,
|
|
"epoch": 0.2071251035625518,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.4382,
|
|
"mean_token_accuracy": 0.07283047214150429,
|
|
"num_tokens": 457356.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.006765127182007,
|
|
"epoch": 0.2112676056338028,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.4372,
|
|
"mean_token_accuracy": 0.07371915206313133,
|
|
"num_tokens": 465928.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.962953233718872,
|
|
"epoch": 0.21541010770505387,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.3086,
|
|
"mean_token_accuracy": 0.07748371884226798,
|
|
"num_tokens": 475269.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.846234321594238,
|
|
"epoch": 0.2195526097763049,
|
|
"grad_norm": 0.89453125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.2837,
|
|
"mean_token_accuracy": 0.07470350190997124,
|
|
"num_tokens": 485450.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.911148262023926,
|
|
"epoch": 0.22369511184755592,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.3174,
|
|
"mean_token_accuracy": 0.07727364040911197,
|
|
"num_tokens": 494029.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.862251615524292,
|
|
"epoch": 0.22783761391880697,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.2856,
|
|
"mean_token_accuracy": 0.07773918211460114,
|
|
"num_tokens": 503253.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.789915609359741,
|
|
"epoch": 0.231980115990058,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.2873,
|
|
"mean_token_accuracy": 0.07239806838333607,
|
|
"num_tokens": 512857.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.920849275588989,
|
|
"epoch": 0.23612261806130902,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.3294,
|
|
"mean_token_accuracy": 0.07429277002811432,
|
|
"num_tokens": 522319.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.767004537582397,
|
|
"epoch": 0.24026512013256007,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.2731,
|
|
"mean_token_accuracy": 0.07617050744593143,
|
|
"num_tokens": 531469.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.817885971069336,
|
|
"epoch": 0.2444076222038111,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.2883,
|
|
"mean_token_accuracy": 0.07853077799081802,
|
|
"num_tokens": 540470.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.739614582061767,
|
|
"epoch": 0.24855012427506215,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.2711,
|
|
"mean_token_accuracy": 0.07688240185379983,
|
|
"num_tokens": 550317.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.804083776473999,
|
|
"epoch": 0.2526926263463132,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.2542,
|
|
"mean_token_accuracy": 0.07694350108504296,
|
|
"num_tokens": 559409.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.721450233459473,
|
|
"epoch": 0.25683512841756423,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.2393,
|
|
"mean_token_accuracy": 0.0897148072719574,
|
|
"num_tokens": 568726.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.8209740161895756,
|
|
"epoch": 0.2609776304888152,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.26,
|
|
"mean_token_accuracy": 0.07505979798734189,
|
|
"num_tokens": 577926.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.7856145858764645,
|
|
"epoch": 0.2651201325600663,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.2553,
|
|
"mean_token_accuracy": 0.07944877073168755,
|
|
"num_tokens": 586695.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.730410480499268,
|
|
"epoch": 0.26926263463131733,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.264,
|
|
"mean_token_accuracy": 0.07737773731350898,
|
|
"num_tokens": 596250.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.764742994308472,
|
|
"epoch": 0.27340513670256833,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.2863,
|
|
"mean_token_accuracy": 0.07472797855734825,
|
|
"num_tokens": 605761.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.720160007476807,
|
|
"epoch": 0.2775476387738194,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.2106,
|
|
"mean_token_accuracy": 0.07840372547507286,
|
|
"num_tokens": 614313.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.699429607391357,
|
|
"epoch": 0.28169014084507044,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.2329,
|
|
"mean_token_accuracy": 0.079537483304739,
|
|
"num_tokens": 624470.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.716359567642212,
|
|
"epoch": 0.28583264291632143,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.1781,
|
|
"mean_token_accuracy": 0.07983956411480904,
|
|
"num_tokens": 633458.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.705728006362915,
|
|
"epoch": 0.2899751449875725,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.1396,
|
|
"mean_token_accuracy": 0.08111847266554832,
|
|
"num_tokens": 643251.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.584441757202148,
|
|
"epoch": 0.29411764705882354,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.2011,
|
|
"mean_token_accuracy": 0.08007878810167313,
|
|
"num_tokens": 652005.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.649393129348755,
|
|
"epoch": 0.2982601491300746,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.0934,
|
|
"mean_token_accuracy": 0.08189089968800545,
|
|
"num_tokens": 661003.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.623857879638672,
|
|
"epoch": 0.3024026512013256,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.0933,
|
|
"mean_token_accuracy": 0.08178795203566551,
|
|
"num_tokens": 670221.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.750538110733032,
|
|
"epoch": 0.30654515327257664,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.3477,
|
|
"mean_token_accuracy": 0.07552824206650258,
|
|
"num_tokens": 680756.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.595584869384766,
|
|
"epoch": 0.3106876553438277,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.0678,
|
|
"mean_token_accuracy": 0.08516154885292053,
|
|
"num_tokens": 689641.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.549710321426391,
|
|
"epoch": 0.3148301574150787,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.995,
|
|
"mean_token_accuracy": 0.0825497955083847,
|
|
"num_tokens": 698209.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.580410575866699,
|
|
"epoch": 0.31897265948632975,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.9957,
|
|
"mean_token_accuracy": 0.08926377072930336,
|
|
"num_tokens": 706712.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.595743894577026,
|
|
"epoch": 0.3231151615575808,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.1863,
|
|
"mean_token_accuracy": 0.08130085095763206,
|
|
"num_tokens": 716087.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.5478174686431885,
|
|
"epoch": 0.3272576636288318,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.0828,
|
|
"mean_token_accuracy": 0.07734477743506432,
|
|
"num_tokens": 725913.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.606314182281494,
|
|
"epoch": 0.33140016570008285,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.1209,
|
|
"mean_token_accuracy": 0.0859277956187725,
|
|
"num_tokens": 735020.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.527130556106568,
|
|
"epoch": 0.3355426677713339,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.0165,
|
|
"mean_token_accuracy": 0.07862741872668266,
|
|
"num_tokens": 744299.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.538954830169677,
|
|
"epoch": 0.3396851698425849,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.9993,
|
|
"mean_token_accuracy": 0.08304714113473892,
|
|
"num_tokens": 753231.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.5157403469085695,
|
|
"epoch": 0.34382767191383595,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.0619,
|
|
"mean_token_accuracy": 0.08629989027976989,
|
|
"num_tokens": 763043.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.627204179763794,
|
|
"epoch": 0.347970173985087,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.0861,
|
|
"mean_token_accuracy": 0.08552133813500404,
|
|
"num_tokens": 771977.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.432950973510742,
|
|
"epoch": 0.352112676056338,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.9878,
|
|
"mean_token_accuracy": 0.08852855414152146,
|
|
"num_tokens": 780285.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.566639614105225,
|
|
"epoch": 0.35625517812758906,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.047,
|
|
"mean_token_accuracy": 0.08173825293779373,
|
|
"num_tokens": 789980.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.483531761169433,
|
|
"epoch": 0.3603976801988401,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.0026,
|
|
"mean_token_accuracy": 0.09233896881341934,
|
|
"num_tokens": 798700.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.38806381225586,
|
|
"epoch": 0.36454018227009116,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.059,
|
|
"mean_token_accuracy": 0.08308355435729027,
|
|
"num_tokens": 808758.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.56745285987854,
|
|
"epoch": 0.36868268434134216,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.0227,
|
|
"mean_token_accuracy": 0.0889634720981121,
|
|
"num_tokens": 817692.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.4513147354125975,
|
|
"epoch": 0.3728251864125932,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 7.074,
|
|
"mean_token_accuracy": 0.09340473040938377,
|
|
"num_tokens": 827052.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.493870782852173,
|
|
"epoch": 0.37696768848384427,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.0443,
|
|
"mean_token_accuracy": 0.09431338384747505,
|
|
"num_tokens": 836442.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.542994403839112,
|
|
"epoch": 0.38111019055509526,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.9475,
|
|
"mean_token_accuracy": 0.08114949613809586,
|
|
"num_tokens": 846441.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.422606658935547,
|
|
"epoch": 0.3852526926263463,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.9848,
|
|
"mean_token_accuracy": 0.09037281647324562,
|
|
"num_tokens": 856108.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.409952163696289,
|
|
"epoch": 0.38939519469759737,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.9198,
|
|
"mean_token_accuracy": 0.08507395833730698,
|
|
"num_tokens": 865788.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.413805341720581,
|
|
"epoch": 0.39353769676884837,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.9738,
|
|
"mean_token_accuracy": 0.0836036428809166,
|
|
"num_tokens": 874833.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.285798835754394,
|
|
"epoch": 0.3976801988400994,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.9254,
|
|
"mean_token_accuracy": 0.08790519461035728,
|
|
"num_tokens": 883685.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.440755939483642,
|
|
"epoch": 0.40182270091135047,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.9147,
|
|
"mean_token_accuracy": 0.08789389207959175,
|
|
"num_tokens": 893908.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.403847885131836,
|
|
"epoch": 0.40596520298260147,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.0129,
|
|
"mean_token_accuracy": 0.08363201320171357,
|
|
"num_tokens": 903081.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.289476585388184,
|
|
"epoch": 0.4101077050538525,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.8789,
|
|
"mean_token_accuracy": 0.08956534340977669,
|
|
"num_tokens": 912515.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.4823840141296385,
|
|
"epoch": 0.4142502071251036,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.8952,
|
|
"mean_token_accuracy": 0.09587915241718292,
|
|
"num_tokens": 921056.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.4142502071251036,
|
|
"eval_entropy": 7.1376285746153485,
|
|
"eval_loss": 7.034562110900879,
|
|
"eval_mean_token_accuracy": 0.08812946460839273,
|
|
"eval_num_tokens": 921056.0,
|
|
"eval_runtime": 2.122,
|
|
"eval_samples_per_second": 1765.767,
|
|
"eval_steps_per_second": 221.015,
|
|
"step": 500
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 12060,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 10,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 204998070435840.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|